GE Vernova and the AI Electricity Challenge: Between Gas Turbines and Blackout Risks

Context

GE Vernova (GEV) has become the key supplier of gas turbines for hyperscalers building their AI data centers (CNBC, 27/06). Meanwhile, the CEO of a major U.S. utility warns in the Financial Times (27/06): electricity supply physically cannot absorb the growth of AI demand at the current pace. These two signals point to a single bottleneck: electricity.

Data

• Global data center demand: ~945 TWh projected for 2026 (IEA, Electricity 2025), including ~380 TWh for the U.S. (≈40% of total)
• GE Vernova: gas turbine order book accelerating since 2025; lead time of 3 to 5 years for new units
• Grid alert: According to the CEO of a major U.S. utility (FT, 27/06), available capacity margins on the U.S. grid are insufficient to absorb projected AI load growth—risk of outages documented by operators
• AI Data Center Moratorium Act (AOC/Sanders, 25/06): would block any new construction without certified electricity supply proof

Analysis

The AI bottleneck is no longer computing power: it’s electricity. GE Vernova plays a structural role because gas turbines combine rapid response (ramping in minutes) and energy density that data centers require. Nuclear (SMRs) and solar/wind PPAs cannot deliver at the required scale in the next 2–3 years. The blackout risk is not hypothetical: U.S. utility executives are publicly warning regulators about the growing mismatch between electricity supply and demand.

Probability-weighted scenarios

• Base case (55%): Manageable grid strain via load shedding and demand response. GEV consolidates market share. Electricity costs rise 15–25% in data center regions by 2027.
• Bear case (25%): Moratorium Act or equivalent adopted. Freeze on U.S. data center construction, relocations to the EU and Asia.
• Bull case (20%): Accelerated SMR deployment (Microsoft/Constellation) resolves the issue by 2028–2029. GEV + AI-exposed utilities = top-tier "pick and shovel" play.

Portfolio implications

GE Vernova, Constellation Energy, Vistra Energy = "AI infrastructure" positioning rooted in the real economy. Utilities exposed to data centers benefit from rare demand growth in a traditionally flat sector. This theme transcends the short cycle of semiconductors.

Risks & blind spots

Regulatory risk (Moratorium Act) underestimated by consensus. Rising electricity prices = drag on hyperscaler profitability if passed on. Dependence on natural gas exposes GEV to institutional ESG constraints and energy price volatility.

To watch